Prognosis Gene Expression Signature Research Articles

Re-evaluation of publicly available gene-expression databases using machine-learning yields a maximum prognostic power in breast cancer

Gene expression signatures refer to patterns of gene activities and are used to classify different types of cancer, determine prognosis, and guide treatment decisions. Advancements in high-throughput technology and machine learning have led to improvements to predict a patient’s prognosis for different cancer phenotypes. However, computational methods for analyzing signatures have not been used to evaluate their prognostic power. Contention remains on the utility of gene expression signatures for prognosis. The prevalent approaches include random signatures, expert knowledge, and machine learning to construct an improved signature. We unify these approaches to evaluate their prognostic power. Re-evaluation of publicly available gene-expression data from 8 databases with 9 machine-learning models revealed previously unreported results. Gene-expression signatures are confirmed to be useful in predicting a patient’s prognosis. Convergent evidence from approx 10,000 signatures implicates a maximum prognostic power. By calculating the concordance index, which measures how well patients with different prognoses can be discriminated, we show that a signature can correctly discriminate patients’ prognoses no more than 80% of the time. Additionally, we show that more than 50% of the potentially available information is still missing at this value. We surmise that an accurate prognosis must incorporate molecular, clinical, histological, and other complementary factors.

Comprehensive Evaluation of Machine Learning Models and Gene Expression Signatures for Prostate Cancer Prognosis Using Large Population Cohorts.

This systematic evaluation of 15 machine learning algorithms and 30 published gene expression signatures for the prognosis of prostate cancer will assist clinical decision-making.

Integrative proteogenomic analyses of human tumours identifies ADNP as a novel oncogenic mediator of cell cycle progression in high-grade serous ovarian cancer with poor prognosis.

BackgroundDespite toxic side effects and limited durable response, the current standard-of-care treatment for high grade serous ovarian cancer (HGSOC) remains platinum/taxane-based chemotherapy. Given that the overall prognosis has not improved drastically over the past several decades, there is a critical need to understand the underlying mechanisms that lead to tumour development and progression.MethodsWe utilized an integrative proteogenomic analysis of HGSOC tumours applying a poor prognosis gene expression signature (PPS) as a conceptual framework to analyse orthogonal genomic and proteomic data from the TCGA (n = 488) and CPTAC (n = 169) studies. Genes identified through in silico analyses were assessed in vitro studies to demonstrate their impact on proliferation and cell cycle progression.FindingsThese analyses identified DNA amplification and overexpression of the transcription factor ADNP (Activity Dependent Neuroprotector Homeobox) in poorly prognostic tumours. Validation studies confirmed the prognostic capacity of ADNP and suggested an oncogenic role for this protein given the association between ADNP expression and pro-proliferative signalling. In vitro studies confirmed ADNP as a novel and essential mediator of cell proliferation through dysregulation of cell cycle checkpoints.InterpretationWe identified ADNP as being amplified and overexpressed in poor prognosis HGSOC in silico analyses and demonstrated that ADNP is a novel and essential oncogene in HGSOC which mediates proliferation through dysregulation of cell cycle checkpoints in vitro.FundingThe National Cancer Institute of the National Institutes of Health, the V Foundation for Cancer Research and the New Jersey Commission for Cancer Research.

Prognostic stratification improvement by integrating ID1/ID3/IGJ gene expression signature and immunophenotypic profile in adult patients with B-ALL

BackgroundSurvival of adults with B-Acute Lymphoblastic Leukemia requires accurate risk stratification of patients in order to provide the appropriate therapy. Contemporary techniques, using clinical and cytogenetic variables are incomplete for prognosis prediction.MethodsTo improve the classification of adult patients diagnosed with B-ALL into prognosis groups, two strategies were examined and combined: the expression of the ID1/ID3/IGJ gene signature by RT-PCR and the immunophenotypic profile of 19 markers proposed in the EuroFlow protocol by Flow Cytometry in bone marrow samples.ResultsBoth techniques were correlated to stratify patients into prognostic groups. An inverse relationship between survival and expression of the three-genes signature was observed and an immunophenotypic profile associated with clinical outcome was identified. Markers CD10 and CD20 were correlated with simultaneous overexpression of ID1, ID3 and IGJ. Patients with simultaneous expression of the poor prognosis gene signature and overexpression of CD10 or CD20, had worse Event Free Survival and Overall Survival than patients who had either the poor prognosis gene expression signature or only CD20 or CD10 overexpressed.ConclusionBy utilizing the combined evaluation of these two immunophenotypic markers along with the poor prognosis gene expression signature, the risk stratification can be significantly strengthened. Further studies including a large number of patients are needed to confirm these findings.

Prognostic value of genomic signatures in metastatic Clear Cell Renal Cell Carcinoma (mRCC) using The Cancer Genome Atlas (TCGA) data.

4560 Background: Gene-expression signatures for prognosis have been reported in localized RCC. The aim of this study was to validate two different signatures; ClearCode34, a 34-gene signature model (Brooks et al. Eur Urol., 2014) and an 8-gene signature model (Choudhury et al. Eur Urol., 2015) in a population of mRCC. Methods: mRCC patients from five institutions who were part of TCGA KIRC were identified and clinical data were retrieved. Normalized RNAseq gene expression data were downloaded from the public TCGA portal. mRCC patients were dichotomized to good and poor prognostic risk groups using the 2 models. Cox proportional hazard regression was used to investigate an association between molecular subtype and overall survival (OS) from first-line therapy. Results: Overall, 56 patients were included in the analysis. Median age was 62 years. The proportions of patients in IMDC prognostic risk groups were 14%, 66%, and 20% for good, intermediate, and poor risk respectively. Fifty two (92.8%) patients rec...

Clinical Value of Prognosis Gene Expression Signatures in Colorectal Cancer: A Systematic Review

IntroductionThe traditional staging system is inadequate to identify those patients with stage II colorectal cancer (CRC) at high risk of recurrence or with stage III CRC at low risk. A number of gene expression signatures to predict CRC prognosis have been proposed, but none is routinely used in the clinic. The aim of this work was to assess the prediction ability and potential clinical usefulness of these signatures in a series of independent datasets.MethodsA literature review identified 31 gene expression signatures that used gene expression data to predict prognosis in CRC tissue. The search was based on the PubMed database and was restricted to papers published from January 2004 to December 2011. Eleven CRC gene expression datasets with outcome information were identified and downloaded from public repositories. Random Forest classifier was used to build predictors from the gene lists. Matthews correlation coefficient was chosen as a measure of classification accuracy and its associated p-value was used to assess association with prognosis. For clinical usefulness evaluation, positive and negative post-tests probabilities were computed in stage II and III samples.ResultsFive gene signatures showed significant association with prognosis and provided reasonable prediction accuracy in their own training datasets. Nevertheless, all signatures showed low reproducibility in independent data. Stratified analyses by stage or microsatellite instability status showed significant association but limited discrimination ability, especially in stage II tumors. From a clinical perspective, the most predictive signatures showed a minor but significant improvement over the classical staging system.ConclusionsThe published signatures show low prediction accuracy but moderate clinical usefulness. Although gene expression data may inform prognosis, better strategies for signature validation are needed to encourage their widespread use in the clinic.

Analysis of a bone metastasis gene expression signature in patients with bone metastasis from solid tumors

Bone is a major target for metastases in the most frequent solid tumors, which result in severe complications and are a major cause of pain. A bone metastasis gene expression signature was identified using human breast cancer cells in a mouse model. The bone metastasis-related genes encode secretory and cell surface proteins implicated in bone-homing (CXCR4), angiogenesis (CTGF and FGF5), invasion (MMP-1 and ADAMTS1), and osteoclast recruitment (IL11). This signature superimposes on the 70-gene poor prognosis gene expression signature for breast cancer, and only ADAMTS1, CTGF and IL11 were found to be overexpressed in human primary breast cancers with bone relapse. We analyzed the expression of the six bone metastasis-related genes in bone metastases from patients with different types of solid tumors, to assess its relevance in human clinical samples. MMP-1, CXCR4, FGF5 and CTGF were found to be overexpressed in tumor cells of human bone metastases when compared to a human normal epithelial cell line. All the analyzed genes were overexpressed in the tumor cells of breast cancer bone metastases when compared to normal breast tissue. We did not detect any differences between the expression of these genes in bone metastases from breast cancer or from other types of solid tumors. Importantly, there was a significant correlation between the expressions of IL11/CTGF, IL11/ADAMTS1, CTGF/CXCR4, CTGF/ADAMTS1, and MMP-1/ADAMTS1, supporting the cooperative function of these proteins in the bone microenvironment, and the potential functional role of these genes in the establishment of bone metastases in vivo.

Use of laser capture microdissection to derive poor prognosis gene expression signatures of epithelial abdominal malignant mesothelioma

e22098 Background: Prior work in abdominal malignant mesothelioma (AMM) has established a poor prognosis profile of biphasic AMM when compared to more favorable prognosis epithelial AMM. It is evident, however, that epithelial AMM shows heterogeneity in biologic behavior and prognosis. Given improvements in therapy and survival in AMM, stratification of patients with epithelial AMM into favorable and poor prognosis subgroups is desirable and gene expression profiles may provide poor prognosis signatures to further guide therapy. Methods: Favorable and poor prognosis subgroups were defined as 1000 days of survival time (based on median survival of 90 patients with AMM at Columbia Presbyterian as of Sept 2007). Fourteen frozen epithelial AMM (7 favorable prognosis, 7 poor prognosis) with high quality RNA were laser capture microdissected using a PALM Zeiss microscope and analyzed using Affymetrix U133 Plus 2 microarrays. Unsupervised hierarchical clustering and supervised subgroup analysis were performed using BRB tools version 3.2. Results: Unsupervised hierarchical clustering revealed 2 main clades that contained 6 of 7 poor prognosis and 6 of 7 favorable prognosis tumors, respectively. Supervised class comparison revealed a 39-gene signature that distinguished the 2 classes (permutation values of p < .001). High expression of Nde1, UHRF1, and EZH2 by qPCR on a set of 32 AMM (using an upper quartile cutoff) was associated with poor survival (Log rank statistic p<.01, p<.003 and p<.03 respectively). Immunohistochemistry performed on tissue microarrays of 122 AMM for p16 loss and increased EZH2, Col3A1 and UHRF1 immunoreactivity was associated with poor survival (Log rank statistic p<.0001, p<.009, p<.0001 and p<.012, respectively). Conclusions: Using a training set of laser captured favorable and poor prognosis epithelial AMM, a 39-gene signature was obtained that is associated with prognosis. Validation of several of the candidate genes by qPCR expression and immunohistochemistry was performed, confirming favorable and poor prognosis strata among the epithelial AMM. Identification of poor prognosis epithelial AMM at the time of initial biopsy/tumor cytoreduction could lead to more aggressive primary therapeutic intervention. No significant financial relationships to disclose.

O-122 Gene expression signatures for prognosis in NSCLC, coupled with signatures of oncogenic pathway deregulation, provide a novel approach for selection of molecular targets

O-122 Gene expression signatures for prognosis in NSCLC, coupled with signatures of oncogenic pathway deregulation, provide a novel approach for selection of molecular targets

Gene expression signatures for prognosis in NSCLC, coupled with signatures of oncogenic pathway deregulation, provide a novel approach for selection of molecular targets

7020 Background: Gene microarray analysis can identify signatures that reflect unique aspects of individual tumors and provide precise prognostic information. Previously, we identified gene express...

Gene Expression Profiling to Predict Outcome of Chemotherapy for Early-Stage Breast Cancer

Rationale Optimal adjuvant systemic treatment of patients with breast cancer following primary surgery is crucial to the achievement of successful outcomes. Although adjuvant chemotherapy and hormonal treatment both increase survival in patients with early-stage breast cancer, it remains difficult to accurately identify which patients require chemotherapy and to select the optimal chemotherapy for individual patients. In general, prognostication relies on broad clinical measures such as tumor size, lymph node status, and hormone receptor status to estimate risk of recurrence and to select treatment. Unfortunately, this leads to homogenous treatment of heterogeneous patients, with a proportion being overtreated and another group undertreated. • With the recent completion of the sequencing of the human genome comes a unique opportunity to advance the means by which clinical decision-making occurs. Through the use of microarray analysis, the expression of a large number of genes can be queried, and gene expression patterns can be used to classify breast cancers into subtypes with varying prognoses and different sensitivities to anticancer agents. • Perou and colleagues were the first group to show that gene expression profiles could be used to stratify patients with early-stage breast cancer into multiple subtypes with markedly different prognoses (P < 0.01).1,2 As an important proof of principle, distinct gene expression profiles have been correlated with varying survival times in different cancer types, including lymphoma3,4 and lung cancer.5 Microarray analysis has also been demonstrated to be more powerful in predicting outcome in patients with stage I or II breast cancer than standard clinicopathologic variables such as lymph node status. In an important pioneering study, the 10year overall survival probability was 55% for patients with the poor prognosis gene expression signature compared with 95% for those with the good prognosis signature.6 • The use of microarray analysis to identify predictive markers requires the ability to correlate differential gene expression with treatment outcome. In the context of neoadjuvant chemotherapy, both of these requirements can be met. Gene expression can be analyzed on the initial tumor biopsy and clinical and pathologic response to chemotherapy can be measured following chemotherapy. Both clinical and pathologic response to neoadjuvant chemotherapy have been shown to be valid predictors of outcome in patients with locally advanced and early-stage breast cancer. Therefore, clinical and pathologic responses can be correlated with breast cancer gene expression profiles in the search for gene expression patterns that predict benefit from chemotherapy. The hope is that gene expression profiles may aid in the determination of which patients will benefit from chemotherapy and which will not.